An in-field automatic fruits recognition and classification system

Fruit is an important part of daily diet around the globe. Automatic fruit classification and recognition is an ill-posed problem. Till now many machines learning models have been developed to classification fruits. However, the classification performance of these techniques is reduced during poor weather and environmental conditions in real-time processing applications. It is quite challenging to automatically classify the fruits from images, when the images are captured from a different viewing angle. This paper proposes an in-field automatic fruit recognition and classification model. Convolution Neural Network (CNN) used to extract image features, Recurrent Neural Network (RNN) used to label and sequence the extracted by CNN. Long-Short Term Memory (LSTM) used to categorize the fruits by using image features that have already been selected and extracted. Experimental and simulation results demonstrate that the proposed system outperforms conventional Machine learning applications. Moreover, the proposed system has been packed into real-time support for the recognition of fruits.